The continued prevalence of nosocomially acquired infections and concomitant clinical and epidemiological interest in typing isolates have resulted in many large clinical laboratories using molecular epidemiological methods to investigate the relatedness of clinically isolated microorganisms. In the Microbiology Service, we are currently evaluating different approaches to molecular strain typing, including plasmid analysis, restriction endonuclease analysis of genomic DNA, pulsed-field gel electrophoresis (PFGE) of genomic DNA, and typing methods based on the polymerase chain reaction. The goal of this project is to develop a coordinated battery of typing methods that can be used to provide information on a wide variety of clinically significant microorganisms. This would permit us to respond promptly in potential outbreak or nosocomial spread situations. Our technical development of these methods includes simplifying procedures so they can be done rapidly, and thereby be of greater epidemiological use. The use of molecular typing methods greatly expands the species of organisms for which we will be able to offer typing. Additional uses of typing availability will be to look at the relatedness of species common to particular patient populations.Thus far, the use of plasmid analysis and PFGE has proven reliable and sufficiently discriminatory for staphylococci. PFGE will also be suitable for a variety of gram-negative rods, such as Serratia. PFGE, however, is slow and labor intensive, often needing to be repeated. We have found that, for some organisms, getting an adequate number of bands for sufficient discrimination is difficult. Our recent work with random amplified polymorphic (RAPD) DNA assays may prove to be of greater use for these organisms, such as has been the case with Roseomonas and Burkholderia species. We have already used our molecular typing procedures to investigate several questions of either patient-to-patient spread, or common source acquisition of pathogens. By working in conjunction with the Hospital Epidemiology Service, we have been able to quickly resolve these issues.Our current developmental concerns are to define optimal procedures for each group of significant pathogens so we can respond reliably when needed. At present we can reliably type strains of staphylococci, enterococci, Pseudomonas, Roseomonas, Serratia, and Burkholderia by either PFGE or RAPD. We are planning to enhance our ability to utilize RAPD in place of PFGE for preliminary results, because RAPD can be done much more rapidly and often more reliably and definitively than PFGE. We completed a large-scale investigation of a lethal Klebsiella outbreak that was seriously compromising the colony of marmosets at the Poolesville veterinary facility. We were able to timely and accurately identify the infecting strain by RAPD's rapid results. This allowed us to recognize the infected marmosets promptly, define a program to reduce or eradicate carriage, and, thereby, bring the outbreak under control. The ease and speed of RAPD was critical to the success of this project, as PFGE was too slow and labor intensive for evaluating hundreds of strains.We are also working on setting up a data repository for PFGE and RAPD patterns to be able to store, retrieve, and manipulate the strain-typing data that we generate. We are using a new software program (Molecular Analyst, Bio-Rad Labs) to do this. Our early evaluation of this program is very encouraging. It will allow us to retrieve hospital pattern data (e.g., for detecting trends in the distribution of antibiotic resistant strains) as well as to follow a particular patient or set of patients in a longitudinal fashion, looking at their colonization/infection patterns. We are particularly interested at looking at specific patient populations that have recurrent infections, to determine the patterns of colonization, infection, and relapse/reinfection that can now be more accurately assessed by these methods.